Skip to content

CODEX_FOR_OSS_APPLICATION.md

Latest commit

 

History

History
83 lines (57 loc) · 3.5 KB

CODEX_FOR_OSS_APPLICATION.md

File metadata and controls

83 lines (57 loc) · 3.5 KB
Raw

Codex for Open Source application prep

Official form: https://openai.com/form/codex-for-oss/

Current repo: https://github.com/Dmatut7/codex-flow Maintainer role: primary maintainer

Official fit

OpenAI says the program is for maintainers of active public OSS projects with meaningful usage, ecosystem importance, or evidence of real maintenance work. Selected maintainers may receive 6 months of ChatGPT Pro, API credits for core OSS work, and possible Codex Security access.

codex-flow should position itself as a Codex maintainer-workflow project: it helps maintainers triage issues, review PRs, investigate bugs, and run release smoke checks in parallel with resumable journals.

Current readiness

  • Public GitHub repository.
  • Public release: https://github.com/Dmatut7/codex-flow/releases/tag/v0.2.3
  • MIT license.
  • Clear README with install and usage path.
  • CLI install path: npm install -g codex-flow. GitHub fallback: npm install -g github:Dmatut7/codex-flow.
  • Codex skill install path: codex-flow install-codex, then users can ask Codex to use a dynamic workflow in any project.
  • Local verification: npm run typecheck, npm test, GitHub Actions CI, and public npm install smoke.
  • Published package readiness: codex-flow is public on npm, root JS import works, root TypeScript declarations work, prepublishOnly gates publish with typecheck + tests.
  • Real Codex smoke verified on 2026-06-05:
    • codex-flow smoke --backend codex-sdk → SMOKE_OK, pong:true, non-zero usage.
    • codex-flow smoke --backend codex-exec → SMOKE_OK, pong:true, non-zero usage.
    • codex-flow smoke --backend openai-responses with no API key → SMOKE_SKIPPED, exit 0.
  • Maintainer workflow examples included: bug investigation, PR review, issue triage, release smoke.
  • Project directly targets maintainer automation, not generic task management.

Remaining weaker signal:

  • Adoption signals are still early: collect stars, issues, outside-user feedback, and usage examples after launch.

Form fields

Email:

Use the email associated with the ChatGPT account.

GitHub username:

Dmatut7

Repository URL:

https://github.com/Dmatut7/codex-flow

Role:

Primary maintainer

OpenAI Organization ID:

Fill from the OpenAI dashboard when applying.

Why does this repository qualify? (max 500 chars)

codex-flow is an open-source workflow orchestrator for Codex. It helps maintainers split complex OSS work into parallel, resumable Codex sub-agents for bug triage, PR review, issue investigation, and release checks. It includes journaling, keyed replay, schema validation, real Codex backends, and a Codex skill so users can trigger workflows with natural language.

How will you use API credits for your project? (max 500 chars)

Use credits to run and test Codex-backed maintainer workflows: PR review fan-out, issue triage, release smoke checks, adapter compatibility tests, and example workflows for OSS users. Credits would also help validate real Codex behavior across codex-sdk, codex-exec, and Responses without pushing cost onto early contributors.

Anything else we should know? (max 500 chars)

This project is built specifically around Codex maintainer automation. It turns Codex App/CLI into a natural-language workflow driver with parallel execution, resumable journals, deterministic replay, soft budgets, and real workspace-write safeguards, so maintainers can use Codex on larger OSS tasks without losing auditability or restartability.